Data acquisition via RS-232, limit on sample rates?

[Marion]

I am communicating with an instrument via RS-232 and need deterministic sampling of the measurement output - I am using software timing to control the interrogation of the instrument and so I'm assuming i'm limited to low sample rates (~1Hz). Are there any alternatives to allow faster sample rates?

[Chris_Rake]

If you are using SW timing you're going to be at the mercy of the operating system. So Windows systems will be very poor while RT systems should offer much more deterministic behavior. I believe you should be able to get much faster than 1 Hz by several orders of magnitude. What type of gitter can you tolerate?

[Marion]

thanks Chris, I am using the data for dynamic modelling and control purposes so i want the time-base to have as little jitter as possible for stability purposes. I'm not too sure what the update rate of the instrument is but based on my knowledge of the process I think an acquisition rate of around 10Hz would be acceptable. What do you suggest?

[Chris_Rake]

Depending on the level of jitter, you might be able to get something that will pass with a high-priority thread in windows. Its worth a shot at least!

[Steven Bird]

Marion, it depends how deterministic the performance you need from your system. If the jitter is acceptable from using RS232 then use it. However these factors really depending on the loop/sampling rate of your application. Bearing in mind that RS232 is a buffered interrupt comms you will be limited by the operating system. If your instrument simply was deteministic in its sampling then sending the data with timestamps over the RS232 link is fine.

Hope this helps.

Steven Bird

[Marion]

Thanks Steven, good point to look at the sampling performance of the instrument itself, at least for purposes of collecting data for modelling. From the point of view of real-time control it sounds RS-232 may be just a bit too unstable - with Windows OS at any rate